In a semiconductor imaging device, a semiconductor substrate forms an active detection medium (detector), which is subdivided into detecting cells. In one form of semiconductor imaging device, for example as described in International patent application WO95/33332, the detector is joined to a readout semiconductor substrate (for example a CMOS readout chip) containing readout cells for reading from individual detector cells in a one-to-one correspondence. The detector and readout chip are flip-chip joined to each other and the readout chip is connected, by further readout electronics, to an analogue to digital converter (ADC) for providing a digitized output from each detector-readout cell.
FIG. 1A of the accompanying drawings illustrates a cross section through a radiation detector 10 for such an imaging device. A metal layer 12 is formed on the radiation entrance surface 17 of the semiconductor substrate 11. The metal layer 12 forms a bias electrode. Metal contacts 13 are formed on the opposite surface 18 of the semiconductor substrate 11 to the radiation entrance surface 17. Each of the metal contacts 13 on the surface 18 define the position of a detector cell, for example a pixel detector, within the radiation detector substrate 11 so that the radiation detector provides a position sensitive device indicating the position at which radiation enters the device.
Specifically, each detector cell provides a current or voltage output having a magnitude and/or duration representative of, and typically substantially proportional to, the magnitude and/or duration, respectively, of incident radiation 19 in the vicinity of the detecting cell. During irradiation, the metal layer 12 on the radiation entrance surface 17 is held at a constant potential (bias voltage), typically a negative potential of a few hundred Volts. Before irradiation, the metal contacts 13 are set at a different potential (for example +5 V), which should ideally be the same for all contacts in order to provide an electric field that is substantially uniform over the volume of the semiconductor substrate 11. During irradiation, the voltage at each contact 13 may decrease (for example to +2 Volts), as a result of which the electric field uniformity is not significantly perturbed.
Such imaging devices have been implemented and tested by the assignee of the current invention using different materials for the semiconductor detector substrate 11, namely silicon (Si) and cadmium zinc telluride (CdZnTe), the latter being a most attractive choice due to its significantly higher sensitivity to X-ray energies over 10 keV for a detector thickness around 1 mm.
A technique for forming metal contacts (e.g. pixels) on the semiconductor substrate 11 is disclosed in patent application No. PCT/EP96/05348 assigned to the assignee of the current invention. According to this technique, as illustrated in FIG. 1B of the accompanying drawings, metal contacts 13 are formed with the surface resistivity of the pixel contact side 18 of the detector kept high by means of a passivation layer 14 between the metal contacts 13. The passivation layer 14 minimizes leakage currents between the metal contacts 13. An example of a material for the passivation layer is aluminium nitride (AlN). Although this technique has provided detectors with high surface resistivities, other problems, not addressed in the prior art, have been observed during tests of the imaging devices.
During laboratory tests, CdZnTe detectors manufactured according to the basic teaching of PCT/EP96/05348 were tested. As a subject for the tests, rather than a single layer of metallization 12 as shown in FIGS. 1A and 1B, a metallization layer was used which comprised two metal layers 15, 16, as shown in FIG. 1C of the accompanying drawings. Each of the metal layers 15 and 16 was 50 nm thick. In some examples, the layer 15 was made of platinum (Pt) due to its better adherence to CdZnTe and the layer 16 of was made of gold (Au). During tests, the detectors 10 exhibited a nonlinear response at conditions of increased exposure (high mAs setting of the X-ray tube) and/or high X-ray tube voltage settings (i.e. high incident X-ray energies). Specifically, images at such conditions have shown a saturation (maximum possible ADC value) of the output given by several detector cells. The number of saturated cell outputs increased as the exposure and/or tube voltage increased. Furthermore, with such conditions, the output values from non-saturated detector cells showed a wider spread around their mean value with a larger tail to their distribution near saturation. Moreover, the saturating cells tended to be located mostly near the detector edges.
Below saturation, the actual ADC value for each cell was proportional to the time integral of the current output of the corresponding detector element. Saturation indicated, therefore, an increased integrated (over time) current output of the detector cells. It was decided to investigate the time dependence of the current output during X-ray exposure by tracing this using an oscilloscope. The 1.5 mm thick detector was biased at -700 V and the X-ray tube was placed 90 cm above.
For exposure and tube voltage settings where the detector response was linear (no saturation problems), the total detector current output, converted into a voltage output by means of an electronic circuit, appeared as a negative square pulse on the oscilloscope. The duration of the pulse was proportional to the exposure time, and its height was proportional to the X-ray intensity incident on the detector. Since the X-ray intensity was constant during exposure, the pulse height was also substantially constant, as expected. Furthermore, the pulse edges were substantially sharp, indicating that the detector response was confined in the time window of the X-ray irradiation. Typical settings leading to this proper pulse shape included tube voltage settings under 60 kV and exposure settings under 5 mAs.
At increased exposure and tube voltage settings, the pulse shape deviated from the proper behaviour described in the previous paragraph:
(i) the pulse height was not stable--most of the time it increased over time indicating an unstable detector response to constant incident X-ray intensity; PA1 (ii) the pulse rising (trailing) edge was not sharp indicating a slow return to zero current output after termination of X-ray irradiation--this rising (in magnitude) current output was compatible with saturation behaviour since the actual output was significantly higher than the output corresponding to incident radiation.
Typical exposure and tube voltage combinations providing such nonlinear detector response were 20 mAs and 100 kV, respectively, but the effect was observed at 1 mAs and 60 kV.
One hypothesis for explaining this nonlinear behaviour assumes a possible elimination or reduction of the Schottky potential barrier in the Schottky contact between the metallization layer and the semiconductor substrate at the radiation entrance surface. This reduction or elimination of the barrier could be caused by an excess accumulation of holes in the vicinity of the contact. Such a hypothesis is compatible with the appearance of non-linearities at large exposures and/or X-ray energies that create a large number of electron hole pairs in the semiconductor volume. The hypothesis is further supported by the fact that the nonlinear behaviour does not appear if the semiconductor material is silicon instead of CdZnTe since the metal contact on Si is formed as a P-N junction and not as a Schottky contact.
Yet further explanation of the effect may take into account the fact that saturation effects occur first for cells nearer to the detector edge. Such an effect may be due to high electric fields in the vicinity of the edges of the metallization layer 12.
It is the aim of the current invention to solve or mitigate the problems described in the previous paragraphs. Specifically, the invention aims at restoring linear detector behaviour for all possible combinations of exposure and X-ray tube voltage.